diff --git a/CHANGELOG.md b/CHANGELOG.md
index 846f045739..e7238b28bb 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -14,6 +14,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 ### Fixed
 
+- Fixed a bug when not explicitly passing `embedding_sizes` to the `TabularClassifier` and `TabularRegressor` tasks ([#1067](https://github.com/PyTorchLightning/lightning-flash/pull/1067))
+
 ### Removed
 
 ## [0.6.0] - 2021-13-12
diff --git a/flash/tabular/classification/model.py b/flash/tabular/classification/model.py
index 392132f4c7..052e59e04a 100644
--- a/flash/tabular/classification/model.py
+++ b/flash/tabular/classification/model.py
@@ -42,7 +42,8 @@ class TabularClassifier(ClassificationTask):
     Args:
         num_features: Number of columns in table (not including target column).
         num_classes: Number of classes to classify.
-        embedding_sizes: List of (num_classes, emb_dim) to form categorical embeddings.
+        embedding_sizes: List of (num_classes, emb_dim) to form categorical embeddings (or ``None`` if there are no
+            categorical fields in the data).
         loss_fn: Loss function for training, defaults to cross entropy.
         optimizer: Optimizer to use for training.
         lr_scheduler: The LR scheduler to use during training.
@@ -63,7 +64,7 @@ def __init__(
         self,
         num_features: int,
         num_classes: int,
-        embedding_sizes: List[Tuple[int, int]] = None,
+        embedding_sizes: Optional[List[Tuple[int, int]]] = None,
         loss_fn: Callable = F.cross_entropy,
         optimizer: OPTIMIZER_TYPE = "Adam",
         lr_scheduler: LR_SCHEDULER_TYPE = None,
@@ -75,7 +76,11 @@ def __init__(
     ):
         self.save_hyperparameters()
 
-        cat_dims, cat_emb_dim = zip(*embedding_sizes) if embedding_sizes else ([], [])
+        if embedding_sizes:
+            cat_dims, cat_emb_dim = zip(*embedding_sizes)
+        else:
+            cat_dims, cat_emb_dim, embedding_sizes = [], [], []
+
         model = TabNet(
             input_dim=num_features,
             output_dim=num_classes,
diff --git a/flash/tabular/regression/model.py b/flash/tabular/regression/model.py
index 18d9fc453f..0ceace0acc 100644
--- a/flash/tabular/regression/model.py
+++ b/flash/tabular/regression/model.py
@@ -40,7 +40,8 @@ class TabularRegressor(RegressionTask):
 
     Args:
         num_features: Number of columns in table (not including target column).
-        embedding_sizes: List of (num_classes, emb_dim) to form categorical embeddings.
+        embedding_sizes: List of (num_classes, emb_dim) to form categorical embeddings (or ``None`` if there are no
+            categorical fields in the data).
         loss_fn: Loss function for training, defaults to cross entropy.
         optimizer: Optimizer to use for training.
         lr_scheduler: The LR scheduler to use during training.
@@ -60,7 +61,7 @@ class TabularRegressor(RegressionTask):
     def __init__(
         self,
         num_features: int,
-        embedding_sizes: List[Tuple[int, int]] = None,
+        embedding_sizes: Optional[List[Tuple[int, int]]] = None,
         loss_fn: Callable = F.mse_loss,
         optimizer: OPTIMIZER_TYPE = "Adam",
         lr_scheduler: LR_SCHEDULER_TYPE = None,
@@ -71,7 +72,11 @@ def __init__(
     ):
         self.save_hyperparameters()
 
-        cat_dims, cat_emb_dim = zip(*embedding_sizes) if embedding_sizes else ([], [])
+        if embedding_sizes:
+            cat_dims, cat_emb_dim = zip(*embedding_sizes)
+        else:
+            cat_dims, cat_emb_dim, embedding_sizes = [], [], []
+
         model = TabNet(
             input_dim=num_features,
             output_dim=1,
diff --git a/tests/tabular/classification/test_model.py b/tests/tabular/classification/test_model.py
index a0af90c924..e411c945e9 100644
--- a/tests/tabular/classification/test_model.py
+++ b/tests/tabular/classification/test_model.py
@@ -67,7 +67,7 @@ def test_init_train_no_num(tmpdir):
 @pytest.mark.skipif(not _TABULAR_TESTING, reason="tabular libraries aren't installed.")
 def test_init_train_no_cat(tmpdir):
     train_dl = torch.utils.data.DataLoader(DummyDataset(num_cat=0), batch_size=16)
-    model = TabularClassifier(num_classes=10, num_features=16, embedding_sizes=[])
+    model = TabularClassifier(num_classes=10, num_features=16)
     trainer = Trainer(default_root_dir=tmpdir, fast_dev_run=True)
     trainer.fit(model, train_dl)